At the county level, the model predicted Romney’s vote share within 5 percent in 71 percent of the counties, and within 10 percent in 94 percent of the counties. This is actually a substantial improvement over the model’s previous performance at the county level, suggesting that the demographic factors driving the election are solidifying even further.

Just to re-emphasize this point: By looking at nothing more than the percentage of Mormons, evangelicals, African-Americans, Latinos, and college-educated voters in counties that voted from South Carolina through Super Tuesday, you could forecast Romney’s vote share within five points in 103 of the 146 counties in Alabama and Mississippi that have returned votes so far. You’d be within 10 points in all but nine. It’s not that great of an exaggeration to say that all the advertising, campaigning, gaffes, and everything else are superfluous to these underlying factors right now…

When I plug numbers into the delegate calculator using my assumptions of how things will play out, it is a very close call as to whether Romney will get to 1,144. If I’m right, this means, literally, that the Republican race could hinge on what the voters in the Republican primaries in Maxine Waters’, Nancy Pelosi’s, and Jose Serrano’s districts do. That can’t be a very comforting thought for the campaigns, or for the parties.